drug-retrosynthesis

What problem does it solve?

Drug retrosynthesis guidance can be slow and uncertain when you need to estimate synthetic accessibility and plausible precursor pathways for a novel small molecule.

Core Features & Use Cases

• Retrosynthetic pathway prediction: Uses IBM RXN’s transformer models to generate ranked precursor trees for a target molecule from its canonical SMILES.
• Synthetic accessibility estimation: Returns confidence scores tied to the feasibility of proposed synthesis routes.
• Multi-step exploration control: Lets you limit search depth via a configurable maximum number of retrosynthetic steps.

Use Case: For an undocumented fluorinated gas analog provided as SMILES, run the skill to obtain ranked retrosynthetic pathways (with confidence) to help prioritize candidate synthesis routes for follow-up research.

Quick Start

Run drug-retrosynthesis by exporting your IBM RXN_API_KEY in your terminal and then executing the evaluate_ibm_rxn.py script with your target molecule SMILES and a small step limit.